?? DIY PS Audio Humbuster

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
By the way, it is very difficult to measure the DC offset on the power line. Most meters will not do it accurately. On a typical Fluke meter, the input circuits would have to be set to accomodate the full waveform, so there is no clipping on the input stage. So for 120VAC, the meter would have to be on the 200V scale. (169V peak on the sine wave) Most 3-1/2 digit meters are limited to 0.1V resolution on this scale. Also the meter specs usually say the last digit can always be off by one. So, unless there is more than 200mV of DC, the result you measure is pretty meaningless. If you allow the meter to go to the 200mV DC scale, then the input stages are being driven into clipping most of the time, and again the reading is meaningless. (it doesn't tell you that... it will give you a number) A 4-1/2 digit meter will start to give meaningful data, but I wouldn't call it accurate.
 
paulmc said:
If you put 1 volt into two diodes in series and a load resistor at the other end, what do you get out?

Nothing.

That's right but connect another pair of diodes in reverse parallel feed it with say 3V AC with a 1V offset and what do you get?

Has the offset dissapeared? Maybe the reduction in hum is due to an increase in mains supply impedance.

You don't have to feed mains into the HB to test if it removes DC offset. The effect will be much easier to see with 10VAC and an offset of 1V.
 
Richard C said:


That's right but connect another pair of diodes in reverse parallel feed it with say 3V AC with a 1V offset and what do you get?


Uh, nothing as well. A single diode going in one direction has a 0.6 volt drop. Put less than that in and you get nothing out. Reverse the diode and it blocks all the voltage. So, either way you put the diode, it will either block 0.6 volts or all of the DC.
 
The diode arrangement I'm refering to is the one you described as being used in the HB. I quote:

"Think of four diodes, two in series going one way and two in series going the other way. Now picture the string of two: one with the cathode at the AC line and the other with the anode at the AC line"

The question was "feed it with say 3V AC with a 1V offset and what do you get?"

You don't get "nothing"

And crucially is the output more or less symetrical?
 
OK, it must be me. I'll try one more time but then I gotta get back to work.

Here's a sketch of a simple parallel string of series diodes that I wrote about. Maybe it's just my lack of communication skills why this wasn't obvious.

When you look at the drawing it'll be obvious to you about blocking DC. This is really simple stuff.

As far as ther assymetry correction, it's the DC that's level shifting the AC. When the DC is gone the AC gets symmetrical.

The trick is getting the AC then to look and act correctly around the zero cross point - where the diodes are not conducting. Our solution is not perfect around this point - it's not bad, but not perfect. The good news is that around the zero cross point you're not really drawing any current, so it's ok.

Cheers! Now back to work for me.

Have fun guys.
 

Attachments

  • diodes.gif
    diodes.gif
    29.1 KB · Views: 1,404
The total amount of DC this simple double-diode-resistor circuit is able to correct is equal to about 1.2V times the relationship between the time the absolute value of the mains voltage is below 1.2V on each cycle and the lenght of a cycle

That approach provides correction for, let's say, up to +-20mV DC that may be enough sometimes but not allways. I've seen up to 500mV DC on my mains AC supply

The full +-1.2V correction range is achieved only by connecting a bipolar capacitor in paralell with the diodes

To measure DC content on any mains line just use a lowpass filter with a reasonably low Fc [ie: 100K and 10 to 100uF], allow it for some time to settle and measure the output of the filter with a simple multimeter. Measuring equimpent is usually not designed to accurately measure 100mV DC on a 640V p-p wavevorm [230V mains]
 
as far as capacitor values ...

... refer to the older discussion with schematic (someone posted the link earlier in this thread).
some folks built the circuit and had success.
i don't remember if anyone actually measured how much DC they had or otherwise determined how much DC they needed to remove.

mlloyd1
 
Greetings Paul McGowan!!!

Don’t let us run you off before I get a chance to applaud your openness and your support of the sometimes brutal world of DIY!!! I’m constantly amazed at how the likes of Dan Wiggins and Nelson Pass make time to support these forums AND run a company (AND have a life?)!

I am friends with one of the tekkies you had in your San Luis Obispo, CA facility and was in Hi-Fi nirvana when I got to tour your shop (please come back!).

-Casey Walsh
 
Paulmc, I understand your diagram very well and it is exactly this arrangement I have been refering to all along.

Clearly for pure DC anything less than +/- 1V cannot pass through but we're not really talking about DC, the only way of seeing DC on an AC waveform is by integrating it.

The point I'm trying to make is that if you apply AC with an offset (the reality), this diode arrangement won't have much effect except for reducing the ampitude of both halves of the waveform by two diode drops, it can't possibly remove the offset because the diodes directly couple the input to the output.
 
Disabled Account
Joined 2003
Richard C said:
The point I'm trying to make is that if you apply AC with an offset (the reality), this diode arrangement won't have much effect except for reducing the ampitude of both halves of the waveform by two diode drops, it can't possibly remove the offset because the diodes directly couple the input to the output.

an A/C with offset is the same as a DC + A/c with no offset.

The cap will function the same in either situation. and diode really doesn't do anything here in terms of removing A/C.
 
millwood said:


an A/C with offset is the same as a DC + A/c with no offset.

The cap will function the same in either situation. and diode really doesn't do anything here in terms of removing A/C.


Paulmc seems to be thinking of simple DC rather than DC which is riding on the back of much larger AC.

The key components in a DC blocker are the capacitors which should be large so that they present a low impedance at mains frequencies, the diodes should rarely conduct if at all.

The claims for the Humbuster state that it's the diodes that remove the DC and this is what puzzled me about it. The capacitors appear to be several thousand times too small to be effective.
 
Hi!
I also thought that the caps are the key components and the diodes are used to clamp the max voltage across the cap.
...which is in heavy mismatch to the statement of paulmc...

On the other hand paulmc gave the technical details of his product
and tried to convey how it works. His behaviour does not look like someone who sells expensive but effectless circuits...

Hm, do we miss out something?
..may be some interaction with the nonlinear inductive
transformer load...?
Anybody out there who tried it?
My mains supply has nearly no Offset, so I cannot test it (less than 5mV, measured behind a passive filter of 200k and 5x10uF MKT, which reduces the AC component to some millivolts and overdrive of DMM is no issue)....

My two fat 1kVA torroids make some noise, not matter if I put 100.000 uF in series or not. Their noise is definitely not related to DC saturation.....

Bye
Markus
 
Perhaps the brief dead time that occurs during the zero-crossing when no diodes are conducting is long enough in duration to allow the core to reset (or at least get much closer to resetting) so that any cumulative bias is reduced or eliminated. Using two diode drops could be the minimum dead-band for a noticeable improvement.

Yes, it is quite possible I have no idea what I just said, but I’m gonna stick my name under it anyway…

-Casey Walsh
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.